"""Compare reference CW descriptors with the repository's compiled descriptor set."""
import argparse
import json
import re
from pathlib import Path
from google.protobuf import descriptor_pb2 as d


def messages(files):
    result = {}
    def visit(rows, prefix):
        for row in rows:
            full = prefix+'.'+row.name
            result[full] = row
            visit(row.nested_type, full)
    for file in files:
        visit(file.message_type, '.'+file.package)
    return result


def main():
    p = argparse.ArgumentParser(description=__doc__)
    p.add_argument('--reference', type=Path, required=True)
    p.add_argument('--current', type=Path, required=True)
    p.add_argument('--commands', type=Path, required=True)
    p.add_argument('--output', type=Path, required=True)
    args = p.parse_args()
    reference = d.FileDescriptorProto.FromString(args.reference.read_bytes())
    current = d.FileDescriptorSet.FromString(args.current.read_bytes())
    fm, cm = messages([reference]), messages(current.file)
    current_names = {int(v):k for k,v in re.findall(r'"([^"]+)":\s*(\d+)',args.commands.read_text())}
    commands = [{'id':v.number,'ffgo_name':v.name[3:],'current_name':current_names.get(v.number)} for enum in reference.enum_type if enum.name == 'CmdGridFightType' for v in enum.value if v.number]
    mapping, conflicts = {}, []
    def match(a,b,seen):
        if (a,b) in seen:
            return True
        seen.add((a,b))
        if a not in fm or b not in cm:
            return False
        xf, yf = ({v.number:v for v in rows.field} for rows in (fm[a],cm[b]))
        if xf.keys()!=yf.keys():
            return False
        for num,u in xf.items():
            v=yf[num]
            if u.type!=v.type or u.label!=v.label or u.HasField('oneof_index')!=v.HasField('oneof_index'):
                return False
            if u.type==u.TYPE_MESSAGE and not match(u.type_name,v.type_name,seen):
                return False
        # Preserve the partition of fields into oneofs, independent of names.
        groups = lambda fields: sorted(sorted(v.number for v in fields if v.HasField('oneof_index') and v.oneof_index==i) for i in set(v.oneof_index for v in fields if v.HasField('oneof_index')))
        return groups(fm[a].field)==groups(cm[b].field)
    for row in commands:
        row['same_name']=row['ffgo_name']==row['current_name']
        a='.'+reference.package+'.'+row['ffgo_name']
        candidates=[k for k in cm if k.rsplit('.',1)[-1]==row['current_name']]
        seen=set()
        row['wire_shape_equal']=len(candidates)==1 and match(a,candidates[0],seen)
        if row['wire_shape_equal']:
            for k,v in seen:
                if k in mapping and mapping[k]!=v:
                    conflicts.append([k,mapping[k],v])
                mapping[k]=v
    args.output.mkdir(parents=True,exist_ok=True)
    provenance={'reference_descriptor_sha256':__import__('hashlib').sha256(args.reference.read_bytes()).hexdigest(),'current_descriptor_sha256':__import__('hashlib').sha256(args.current.read_bytes()).hexdigest(),'limitations':['Equal wire shapes do not establish enum meanings, field semantics, runtime success, or permission to register numeric aliases.']}
    (args.output/'cw-command-comparison.json').write_text(json.dumps({'provenance':provenance,'commands':commands},indent=2))
    (args.output/'cw-message-mapping.json').write_text(json.dumps({'provenance':provenance,'messages':mapping,'conflicts':conflicts},indent=2))
    print(json.dumps({'commands':len(commands),'same_names':sum(x['same_name'] for x in commands),'equal_wire_shapes':sum(x['wire_shape_equal'] for x in commands),'mapped_messages':len(mapping),'conflicts':len(conflicts),'shape_differences':[x for x in commands if not x['wire_shape_equal']]}))


if __name__=='__main__':
    main()